My experience with EMC immunity is such that I never (now) think about trying to removing a ground loop by connecting a capacitor between XLR pin 1 & chassis. The problem is that the inductance of the leads, either side of the actual capacitor element, tends to be too large at frequencies above 10kHz, to be an effective solution*. Thus, the technique only ever removes the low frequency part of the ground loop, leaving the high frequency part (still containing much RFI), in tact.
After 15 years of designing audio systems with EMC techniques, I now realize that ground loops are (mostly) good, and the best way to reduce RFI, is to design the product using the advantages of ground loops – ie. knowing where they are!
Put simply: the best way to achieve good immunity, is stop RFI from being injected into the active circuit common conductor (inaccurately called “ground”). This is because, once RFI has entered the active circuit, the electronics will demodulate the interference, and we will hear this as an audible noise at the output of the circuit.
* = it is possible to minimize the capacitor lead inductance by using an annular feed-through type of capacitor, but this is rather more expensive and still less effective compared to simply bonding XLR pin 1 to the chassis (see my 2002 paper on the subject).